4.7Ghz its about freaking time!

RainMotorsports

Leadership
Leadership
Joined
Mar 6, 2011
Messages
8,654
Yeah my dreams of 5Ghz might still be impossible with this sample but progress has been made.

Did a bios update and to my suprise the once unbootable 4.6ghz booted and even the unpostable 4.7ghz booted hell 4.8 even posts now. Takes a tiny bit more voltage than one would like on air and its not stable yet but there is light beyond 4.5Ghz for a change.

CPUZ:
47.png

At least 4.5Ghz should be a bit stabler at a bit less voltage than before as well.
 
Not to provide a snicker, the i7 coming later this month is reported to be able to hit 4.5-6 ...stable. :)
 
Not to provide a snicker, the i7 coming later this month is reported to be able to hit 4.5-6 ...stable. :)

You just mean the average yield will right? I know people with 5Ghz+ samples of the current gen on air or water. There is a guy as well over on OCN thats 24/7 tested 5.7Ghz at a rather suicidal voltage and lived to tell the tale using water. There was a couple people that did nothing but buy and test and resell processors garuntee'd to hit particular multipliers on decent board/power supplies. I mean if you want to cheat luck lol.
I was planning on going i7-3770K on the P8Z77-V Pro but that hinged on selling this setup and having the cash for the upgrade.

Processor is insured so i don't mind messing about still.


Whats wrong you still mad that whoever it was couldn't help you?
 
your memory is scary

LOL I only forget who it was you posted a thread like "rain help me". Your machine is a different beast altogether and besides plenty of trial and error to it sometimes. Takes a person to be at the machine that gets what the next move is.
 


But then again, my 2600K had a faulty motherboard--so I doubt it could have done either 24/7.
 
i hit 4.5 quite a while back with my i7 920 air cooled and stable..... gonna get a new chip though 920 is getting wrinkles
 
You just mean the average yield will right? I know people with 5Ghz+ samples of the current gen on air or water. There is a guy as well over on OCN thats 24/7 tested 5.7Ghz at a rather suicidal voltage and lived to tell the tale using water. There was a couple people that did nothing but buy and test and resell processors garuntee'd to hit particular multipliers on decent board/power supplies. I mean if you want to cheat luck lol.
I was planning on going i7-3770K on the P8Z77-V Pro but that hinged on selling this setup and having the cash for the upgrade.

Processor is insured so i don't mind messing about still.

Basically, Intel has reduced the thermal footprint which allowed for higher overclock. I remember the topic you, way back, that you started(5.0 or bust). I'm quite curious as to what I can reach with a closed loop water cooling that I got. Can't be any worse than the almost-burst-into-flames Phenom I that I have. I amazed at the heat that thing puts out. LOL
 
But then again, my 2600K had a faulty motherboard--so I doubt it could have done either 24/7.

You also said you were not able to boot it. But faulty board and all and being a gigabyte. But the fact that you were able to change and hold it post boot at such a low voltage long enough to get a screen shot showed some promise i think.

Basically, Intel has reduced the thermal footprint which allowed for higher overclock. I remember the topic you, way back, that you started(5.0 or bust). I'm quite curious as to what I can reach with a closed loop water cooling that I got. Can't be any worse than the almost-burst-into-flames Phenom I that I have. I amazed at the heat that thing puts out. LOL

Part of the problem isn't even initial heat per say. On my sample the problem is going to be the voltage it takes to run a particular frequency and the only way its going to survive those higher voltages at load is of course serious cooling. But sadly the cooling will barely change the voltage needed since I can't hold the clocks at a low temp as it is.

I have reached the point of diminishing returns with this sample. Its gonna take alot more voltage every 100mhz than it did before. Fact is this chip will run 4.3ghz without even touching the voltage because the stock offset curve puts it at the perfect voltage for 4.3. But the curve gets pretty steep after that for this pos.

Id like to think if i ever have to turn it in on insurance that my next sample will be better. Otherwise I will be on 5Ghz when everyone and their mother is but those architectures will do some awesome shit at that speed anyways.
 
You also said you were not able to boot it. But faulty board and all and being a gigabyte. But the fact that you were able to change and hold it post boot at such a low voltage long enough to get a screen shot showed some promise i think.
The fault was in the board, not just being a Gigabyte. I still think it was one of the best equipped boards of the generation--at least from an engineering perspective. If I was building now, I'd look at a Biostar TZ77XE4.

But yeah, I couldn't boot it at 50x with default settings. I had to use software to up the multi. The voltage was more than 1.05v--that was a CPU-Z display glitch. I don't remember what it actually ran--I think 1.35V. I could boot fine if I disabled the USB 3.0 & GSATA controllers but, with those controllers enabled, I couldn't even boot at stock. Further, turbo multis weren't working--the chip ran at 3.4 or 3.5 when at stock settings. I don't remember if 4.8 was stable--I sent it back after a few days so I didn't have that much time to both enjoy it and maximize my settings. But if memory serves, 4.8 was damn near stable.

All in all, I've been happier with my new-to-me Nehalems. No problems setting them up or running them. The only remaining issue is how DDR3 undervolting goes by 0.1V increments--so my LV RAM runs at 1.4V instead of 1.35V. Other than that, I couldn't be happier with $200 CPUs & boards. Hell of a lot faster than Core i3's and H67 boards.
 
Corespeed.jpgCorespeed2.jpg

Some one explain to me why my Core speed is showing so low, and then it jumps up for a second then back down. Is this a problem? Rain where you at bro?

**** EDIT **** 30 minutes later:

Ok now I have used an ASUS utility from my MB DVD and followed this path of selections ASUS Turbo EVO - Auto Tuning - Auto System Level Up - Automate Over Clocking, bringing me to 2 options: FAST or EXTREME. I chose FAST and my computer went through a series of quick reboots for about 3 times and then booted and I am OC by 31% at the following numbers. But they are still bouncing from a low CPU speed to a peak CPU speed while I am at idle. Is this normal? I ran an XS BENCH test in RealTemp to get the last screenshot at a 26% load. Thanks for the help.

Corespeed3.jpgCorespeed5.jpg
 
Duke don't take this the wrong way but it kills me to see this question about 10 times a day from other people. Its how the processor is designed. It underclocks when not under load to save energy, reduce heat and increase life span. What kills me more is the people who are so afraid of having their computer run slow they want to disable it. Now on crazy overclocks it sometimes becomes necessary to do so but other than that your processor starts to do a damn thing at all the speed jumps back up.

Careful with that auto tuning bullshit ive read some stories about it setting unsafe voltages and crap. (1.38 is considered generally safe for Sandy Bridge on decent air cooling).

If your chip is stable at 0.944V which is damned low but not unheard of. Buddy do you have some room to go. But id test the stability of that overclock first.

Best to learn how to do it yourself and when I have time i will help you others here that can as well. You can already see something stupid Turbo EVO did. There was no reason for it to increase BCLK to 103mhz. Your processor speed is BCLK * Multiplier. Your chips unlocked so all you need to do is increase the multiplier. No real reason to change the BCLK on this chip until you hit about 5.7Ghz and there is no more multipliers to choose from. Oh and if your running that fast we need to get you some good cooling lmao.
 
Coretemp01.jpg

Thanks Rain. I just never noticed my CPU running less then 3000k + at any time in the past. I'm glad that it is normal. I played a game of BF3 and you can see my max temps. What do you think?
 
Plenty good. Lower than mine and its to be expected at such a low voltage. I am assuming thats with the overclock? A quick but not thorough test of stability would be to run through Intel Burn Test. Its easy enough to pass this but any sort of major instability should be revealed by it. IBT - http://www.techspot.com/downloads/4965-intelburntest.html

A better test would be running Prime95 for 12 to 24 hours. You can find alot of information in this sandy bridge stable club thread because they require the stability testing to join - [Official] The Sandy Stable Club **Guides, Voltages, Temps & BIOS Templates** Inc SPREADSHEET

During either of these tests expect temperatures to get artificially high. Please tell me your not on the stock intel cooler :p
 
Very good then. It doesn't qualify as what we would consider good water cooling but its better than most air and certainly way better than stock. I will write out the basics as they apply to Sandy Bridge. Don't have alot of time left in the day have to go to work soon. EDIT - As far as the test goes looks good no immediate signs of instability. But as i said Prime 95 12 to 24 hours only way to be sure. You will blue screen of death if its unstable.

Basic concept of overclocking
Bus Speed times Multiplier = clock speed. Modify either one the clock generator will try and switch the transistors at a faster rate. Back in the day on locked chips we had to adjust the front side bus which also affect ram speed and added 2 other points of instability, the ram and the chipset. Neither is the case with this setup. You actually cant even change the BCLK very far on your processor and there is no need to since your chip has an unlocked multiplier.

For any given architecture and given sample of a chip the transistors will only switch so fast (clock speed) at a particular voltage before becoming unstable. This is remedied by increasing the voltage. The voltage needed is specific to every single chip made even off the same wafer of silicon your individual chip has its own unique properties.

Now overclocking in general is safe given the cooling needs are dealt with. But voltage is where things get hairy. Increasing voltage brings an a much higher increase and heat. Also think of the paths and circuits carved into the silicon as wires. A wire of a given specification can only carry so much current at a particular temperature. Too much voltage and you will fry it, not enough cooling you fry it.

Turbo EVO
Turbo EVO brings some nice easy control into manually changing settings. But you should really get into the bios and do this one step at a time with full control at your finger tips. But when going too far get ready to reset that CMOS.

Voltage Control, Stability
You have 2 options for increasing voltage. There is offset voltage and specified. Normally the processor/motherboard will increase voltage with clock speed. If the chip needs more voltage you can add an offset say 0.05 or 0.10 over. Or you can just specify the voltage manually say 1.00, 1.10 whatever just careful not to go over 1.38 unless your ready to take some REAL risks.

Your motherboard also gives you control over the voltage regulation. 2 aspects to this is the VRM frequency and Load Line Calibration. The cleaner the power and the less ripple in voltage the more stable the overclock. As well as under load you have voltage drop and this is countered with the Load line calibration. Mind blown? Its easier than you think. Higher LLC will help counter the voltage drop Ultra High is fine on this board no more should be needed. VRM frequency is 300 default, 350 is probably a good place when you start needing it i run 370 because i found it suited my chip.

The Rest
Alright so change that BCLK back to 100 and if you want 4.3 Ghz change the multiplier to 43. (If the options you need don't appear in the bios change the mode, I forget which ones there are I have myne set to XMP but there is one other thats fine).
Test your stability with each change, you might wish to jump forward and see at what point where obviously you need more voltage. If you want to go as high as you can without an increase then just work your way backwards, high point and back it down one multiplier at a time. But from what I have seen if your stable at that voltage at 4.3 Ghz you can easily hit 4.5+ with some more voltage.

If the machine refuses to post you've gone too far and it will require resetting the CMOS. This can be quite frustrating and even having wired a switch to the jumper contacts its still a pain in my ass.

Sandy Bridge and PLL Voltage
Sandy Bridge has an interesting quality. Reducing the PLL voltage actually increases stability at high overclocks. The exact number varies, I use 1.45 another guy uses 1.55 this is all down from 1.7. Should only need to change it if your not getting as far as you want to on other settings. Disabling the Internal PLL Overvoltage can help as well.

Quite a few people here can help you, OCN is a good place to go. As with anything don't so anything your not comfortable with. Once again if what you have is actually stable or only needs a little adjustment to become stable you shouldnt have to fiddle much unless your shooting for something much higher. There becomes a point where every little thing is needed. For me thats beyond 4.3 Ghz. For you so far you haven't needed much.

Overclocking insurance can be bought from Intel for the 2500K for 20 bucks. - http://click.intel.com/tuningplan/
 
Ok Rain. I copied all these instructions and will print them. Thanks again for the help. One thing I will try right now for sure is to reduce the BCLK and increase the Multiplier to 43.
 
Back
Top